What is our standard convention for handling NaN values in data streams?

[drslebedev]

Description

Currently, there is no clear convention for handling NaN values in data streams within the gr4 project. This ambiguity could lead to inconsistent behavior across various blocks, resulting in errors or unexpected outcomes.

Some blocks may introduce NaN values due to invalid math calculations. We need to decide how NaN values should be handled. For example, ignored, propagated, or replaced. Note that there are also potential performance implications if multiple blocks must check for NaN.

Considerations

• Do we allow blocks to publish NaN values, or should they filter them out?
• What performance trade-offs are acceptable if every block checks for NaN?

Proposed Actions

1. Establish a clear convention for the usage and interpretation of NaN values in data streams.
2. Document the agreed-upon convention in the developer guidelines.
3. Ensure all blocks implement the same approach to handling NaN values.